Language impairments are common in affective and psychotic disorders, yet their patterns and underlying pathomechanisms remain insufficiently understood. A transdiagnostic perspective provides a framework for identifying shared and disorder-specific language alterations across diagnostic boundaries. Combining natural language processing (NLP) with network analysis enables the investigation of complex associations between linguistic, cognitive, and psychopathological features.

Spontaneous speech from N = 372 participants (119 MDD, 27 BD, 48 SSD and 178 HC) was elicited using four Thematic Apperception Test pictures (~12 min per participant). NLP models were applied to extract latent linguistic variables across various levels, including lexical diversity, syntactic complexity, semantic coherence, and disfluencies. Network analysis was used to relate linguistic variables, psychopathology (SAPS, SANS, HAM-A, HAM-D, YMRS, TLI, GAF), and cognitive performance (attention, verbal memory, recognition, and verbal fluency).

Linguistic variables formed the densest network cluster, with type–token ratio, mean length of utterance, and syntactic complexity emerging as central nodes. Psychopathology variables were less cohesive, while TLI “Impoverishment”, coherence mean, and executive functioning bridged linguistic, cognitive, and psychopathological domains. Network comparison tests revealed no significant differences in linguistic–cognitive network structure across HC, MDD, BD, and SSD.

Linguistic networks show high structural consistency across healthy individuals and patients, whereas psychopathological symptom networks reflect transdiagnostic profiles. These findings support a dimensional and transdiagnostic framework underscore shared language–cognition mechanisms, and highlight executive functioning as key cross-domain connection, which opens up new avenues for dimensional research into the pathophysiological and etiological mechanisms underlying language dysfunctions.

Epidemiological evidence on the incidence and remission of anxiety and depressive disorders is limited. We estimated age- and sex-specific incidence and remission rates of moderate-to-severe anxiety and depressive symptoms using the illness-death model.

The German National Cohort (NAKO) is a cohort of over 200,000 participants aged 19–74 at baseline. Prevalence of probable cases, estimated with the Generalized Anxiety Disorder Scale and the Patient Health Questionnaire data 2014–2019 across five regions, was related to general mortality rates and disorder-specific mortality rate ratios in the illness-death model. The partial derivative of prevalence was modeled as a function of incidence and remission, with parameters estimated via least-squares optimization through 2,000 bootstrap resamples.

The highest incidence rates (per 1,000 person-years) occurred at ages 19–21 for anxiety symptoms: 4.07 (95% CI: 0.00–7.57) in women and 2.55 (0.00–4.94) in men; and at ages 28–34 for depressive symptoms: 4.41 (0.00–9.81) in women and 3.30 (0.00–7.34) in men, all in Hamburg. Remission rates (per 100 person-years) were highest at older ages. For anxiety symptoms, rates peaked at 71.8 years in women (4.10 [0.00–11.94]) and 64.2 years in men (3.00 [0.00–9.23]) in Freiburg. For depressive symptoms, the highest observed was at 74.0 years, both among women (6.61 [0.00–15.50] in Münster) and men (3.58 [0.00–11.51] in Berlin).

Incidence and remission rates of anxiety and depressive symptoms can be estimated from prevalence and mortality data, revealing regional, sex-, and age-related variation. Validation with longitudinal data is warranted.

Ketamine and esketamine produce rapid and sustained antidepressant effects in persons with treatment-resistant depression (TRD). Although it is posited that these effects are largely attributed to N-methyl-D-aspartate receptor antagonism, the potential involvement of the opioid system remains unclear. This systematic review investigates whether ketamine and esketamine antidepressant efficacy is mediated through the opioid system.

We conducted a systematic search of preclinical and clinical studies investigating the potential involvement of the opioid system in the antidepressant effects of ketamine and esketamine. Database searches on PubMed, Cochrane Library, Embase and PsycINFO occurred from inception to September 27, 2025.

16 studies were identified: 12 clinical (n = 790) and 4 preclinical studies. Clinical designs included randomized controlled trials, case reports, pre-post studies and observational cohort studies. Preclinical studies utilized animal models of depression. Only one study examined esketamine. Naltrexone (nonselective opioid antagonist) attenuated ketamine’s effects in three studies, while four reported no such effect and one reported mixed evidence. Genetic markers of opioid receptor subtypes (i.e., OPRM1 and OPRD1) were examined in three studies, but results were inconclusive, potentially due to limited evidence. Separately, opioid use was not associated with ketamine response. Few studies directly examined opioid receptor subtypes.

The reported mixed findings suggest that the opioid system may exert a partial mediating effect of ketamine in TRD. However, given the inconsistent attenuation of ketamine’s antidepressant effects by opioid receptor antagonists, the opioid system likely functions as a context-dependent modulator rather than a primary mediator, particularly at standard antidepressant doses.

Volumetric changes in the superior temporal gyrus and anterior cingulate cortex have been repeatedly reported in studies on schizophrenia. Tractography and functional magnetic resonance imaging studies have suggested that alterations in connectivity involving the superior temporal gyrus and the anterior cingulate cortex are relevant to psychotic symptoms of schizophrenia.

We analyzed nanometer-scale three-dimensional structures of brain tissues of the superior temporal gyrus and the anterior cingulate cortex in eight schizophrenia and eight control cases and evaluated structural parameters of their neurons. We then examined the relation between the neuronal parameters and clinical information including auditory hallucination score.

The obtained results indicated that 1) neurites become thin and tortuous in schizophrenia and that 2) somata become small in schizophrenia. The frequency distribution of neurite curvatures had a broad profile in the schizophrenia cases, whereas the control cases showed sharp peaks. In the scatter diagram of the standard deviation of neurite curvatures, the schizophrenia and control cases formed separate clusters, indicating that all 16 cases analyzed in this study can be assigned to either the schizophrenia or control group simply by using the diagram. The cingulate/temporal ratio of the standard deviation of neurite curvatures showed a strong positive correlation with the auditory hallucination score.

The structural alteration of neurites observed in the schizophrenia cases should influence the function of affected brain areas by hindering communication between distant neurons. We suggest that the interplay of the temporal and cingulate cortices in the whole-brain network is relevant to auditory hallucination.

Bipolar disorder (BD) involves immune-inflammatory dysregulation. This systematic review and meta-analysis assessed complete blood count-based inflammatory indices – neutrophil-to-lymphocyte (NLR), monocyte-to-lymphocyte (MLR), and platelet-to-lymphocyte (PLR) ratios – in BD versus healthy controls (HCs), major depressive disorder (MDD), and across BD mood states.

Databases were searched through June 2025 for observational studies reporting at least one ratio in adults with BD and including as comparators either HCs, MDD, or within-BD mood-state contrasts (mania, bipolar depression, euthymia). Quality was appraised using BIOCROSS. Random-effects meta-analyses, sensitivity analyses, and meta-regressions were performed. GRADE was adapted to rate evidence certainty.

Fifty-one studies (38,309 participants) met the inclusion criteria. Compared to HCs, BD showed higher NLR (SMD = 0.44, p < 0.001) and MLR (SMD = 0.28, p < 0.001). In mania, NLR (SMD = 0.62, p < 0.001), MLR (SMD = 0.51, p < 0.001), and PLR (SMD = 0.18, p = 0.014) were all elevated versus HCs. Depression showed lower PLR (SMD = –0.14, p < 0.001) and euthymia higher NLR (SMD = 0.37, p = 0.002). Compared to MDD, BD had higher NLR (SMD = 0.21, p < 0.001) and MLR (SMD = 0.18, p < 0.001). Similarly, mania showed higher NLR (SMD = 0.53, p < 0.001) and MLR (SMD = 0.41, p < 0.001), while bipolar depression lower PLR (SMD = –0.15, p < 0.001). Mania had higher NLR (SMD = 0.32, p < 0.001), MLR (SMD = 0.32, p < 0.001), and PLR (SMD = 0.14, p = 0.028) than depression and higher MLR than euthymia (SMD = 0.44, p = 0.027), while depression had lower NLR (SMD = –0.28, p = 0.012) and PLR (SMD = –0.22, p < 0.001). Evidence certainty was mixed.

NLR, MLR, and PLR emerge as non-specific, group-level correlates of immune-inflammatory dysregulation in BD, however offering limited discrimination between bipolar and unipolar depression. Notwithstanding their potential role as trait- and state-related markers in BD, further studies are needed to support translation into clinically useful biomarkers.

In the last decade, UN bodies and the WHO call for the abolition of coercion in psychiatry. Studies provide some evidence for interventions to reduce the use of coercion, but it is unclear whether the use of coercion is decreasing in real-world practice. The aim of this study was to gather longitudinal ecologic data on the use of coercive interventions in European countries and to depict trends over time.

For each country, inclusion required access to ecologic datasets spanning a minimum of four years, pertain to a defined population (country or federal state level), and allowing the necessary elements to calculate both the proportion of psychiatric admissions affected by involuntary admissions (IAs) and coercive measures (CMs) and the rate per 100,000 inhabitants. Country experts were accessed via a European network of experts (FOSTREN group).

Data were obtained from Austria, England, Germany, Norway, Sweden, and Switzerland, and covered periods between 4 and 10 years. In no country, an absolute decrease in IA and the use of CM could be observed. Rates of IA per 100,000 inhabitants changed between −5.4% (Sweden) and +37% (Germany). Rates of admissions exposed to any kind of CM changed between +11% (Austria) and +86% (Norway).

The findings suggest a persistence or rise in coercive practices despite national and international policy commitments. An increase in involuntary admissions suggests reasons outside psychiatric hospitals, whereas a disproportionate increase in coercive measures may indicate a change of practice in in-patient psychiatry. Further research is needed to explore the reasons from clinical and societal perspectives.

Considerable effort has been devoted to investigate the neuroimaging correlates and predictors of antidepressant response to ketamine, yet inconsistency in the location and nature of the regional brain effects makes it difficult to unify this research. Despite the revolutionary notion that psychiatric therapeutics show network-level brain representations, investigations into network localization of brain functional effects of ketamine treatment are still lacking.

We initially identified the locations of longitudinal brain functional alterations (increase and decrease separately) induced by ketamine treatment from 16 published studies with 508 depressed patients. By integrating these affected brain locations with large-scale functional MRI datasets from 1113 healthy and 255 depressed individuals, we then leveraged a novel functional connectivity network mapping approach to construct ketamine-induced hyper-functional and hypo-functional networks respectively.

The hyper-functional network mainly involved the subcortical (caudate nucleus and thalamus) and default (medial prefrontal cortex) networks, while its hypo-functional counterpart predominantly implicated the limbic (temporal pole), subcortical (hippocampus and amygdala), and default (lateral temporal cortex) networks.

Our findings may shed light on the neurobiological effects of ketamine from a network perspective, which might represent a crucial step toward fostering the clinical application of ketamine in antidepressant treatment.

Psychiatric comorbidities are common in first-episode psychosis (FEP) and hinder recovery. Problem gambling (PBG), despite potentially serious clinical consequences, remains under-investigated in this population. This study aimed to estimate the incidence of PBG in FEP and identify predictive factors.

This prospective cohort study was conducted at two FEP programmes in Quebec, Canada. Individuals aged 18–35 years diagnosed with FEP between November-1-2019 and January-31-2023 were screened for PBG using the Problem Gambling Severity Index through May-1-2023. The primary outcome was incident PBG. Time-varying Cox regression models were used to estimate hazard ratios (HRs) for candidate predictors.

Among 520 individuals without prior PBG (mean age = 24.6±4.0 years; 28.8% women), 18 developed PBG during a mean follow-up of 478 days, yielding an incidence rate of 2.6 cases/ 100 person-years. In site-adjusted analyses, white ethnicity (HR = 9.7; 95%CI = 1.3–74.8), incomplete high school education (HR = 2.8; 95%CI = 1.1–7.2), stimulant use disorder (HR=2.8; 95%CI = 1.0–7.3), use of D2/D3-5-HT1A partial agonists (HR = 4.6; 95%CI = 1.5–14.1), and prior non-problematic gambling (HR = 3.1; 95%CI = 1.1–8.4) predicted increased risk. Thirteen cases occurred during aripiprazole treatment, which remained associated with increased PBG risk after multivariable adjustment (adjusted HR = 4.7; 95%CI = 1.6–13.9).

Despite the limited number of incident cases, these results suggest that PBG is relatively common PBG and is associated with potential risk factors, including white ethnicity, incomplete high school education, stimulant use disorder, prior non-problematic gambling, and treatment with D2/D3-5-HT1A partial agonists, particularly aripiprazole. These findings underscore the importance of routine screening for PBG and risk-informed antipsychotic prescribing in FEP.

Suicide is a major public health challenge requiring early detection of suicidal ideation (SI). Traditional direct questioning methods suffer from stigma and disclosure bias, failing to identify many at-risk individuals. While machine learning (ML) models show promise, most lack external validation. Indirect screening, using psychosocial data rather than direct SI questions, offers a scalable alternative. This study aimed to externally validate an indirect, ML-based SI screening tool. We tested if a model trained on a Slovenian general population sample retained predictive accuracy when applied to an independent Croatian sample during a period of societal stress (pandemic and earthquakes), assessing performance across age and gender subgroups.

A logistic regression model was trained on a Slovenian sample (N = 2,989) and validated on a Croatian sample (N = 2,364). The model used only indirect predictors, including sociodemographics, life satisfaction, behavioral changes, and Brief COPE subscales. The target outcome was the presence of SI (SIDAS score > 0). Performance was measured by the area under the receiver operating characteristic curve (AUROC).

The model demonstrated strong external validity on the entire Croatian sample, achieving an AUROC of 0.80. Performance remained robust across subgroups: males (AUROC = 0.83), females (AUROC = 0.79), younger adults (AUROC = 0.77), and older adults (AUROC = 0.81). Self-blame, behavioral disengagement, and relationship dissatisfaction were key predictors.

An indirect, ML-based screening tool can reliably identify SI risk in the general population. The model demonstrated strong cross-national transferability and resilience during a societal crisis, proving it is a feasible and valid strategy for population-level prevention.

Psychedelic-assisted psychotherapy has shown potential for psychiatric disorders. However, the magnitude of symptom change in control groups remains poorly understood. We aim to evaluate within-group effects in control groups and compare them to treatment groups in psychedelic trials.

A systematic search was conducted up to 1 July 2025. The study protocol was preregistered in PROSPERO (CRD420251111853).

Fourteen randomized controlled trials (n = 643) were included. Direct between-arm meta-analyses showed greater symptom reductions in treatment compared with control across outcomes, including depressive symptoms (number of study arms [k] = 13; SMD = −0.82; 95% CI = −1.17, −0.47; I2 = 60.1%), posttraumatic stress disorder (PTSD) symptoms (k = 10; SMD = −0.89; 95% CI = −1.14, −0.65; I2 = 0%), and anxiety symptoms (k = 5; SMD = −0.66; 95% CI = −0.94, −0.38; I2 = 0%). Subgroup analyses showed limited evidence that effects differed by placebo type for depressive or PTSD symptoms. Descriptive within-group analyses indicated symptom reductions from baseline in both control and treatment groups, with larger within-group improvements observed in treatment groups across outcomes; notably, larger within-group reductions in PTSD symptoms were observed in inactive placebo groups. Sensitivity analyses showed consistent results.

Control groups in psychedelic trials demonstrated substantial symptom improvement, which may reflect non-specific trial factors (including expectancy and concurrent psychotherapy). These findings emphasize the importance of robust control conditions in psychedelic research and the need for nuanced interpretation of treatment effects.

As in other European countries, mental healthcare in Belgium has to deal with the increasing cultural diversity that exists within society. However, commitment of the Belgian healthcare system toward cultural diversity remains weak, and clear guidelines on culturally competent psychiatric practice are still lacking.

Three focus groups with professional caregivers, three with adult patients, and one with young adults in the transition age were organized. The seven focus groups each consisted of 5–10 participants. Two brainstorming sessions with a total of 15 experts were organized a priori to delineate focus group topics. Data analysis software MAXQDA 24 was used for thematic analysis.

The thematic tree consists of the central theme “culturally sensitive mental healthcare” with five main themes (i.e., vulnerable population, language barrier, mental healthcare stigma, spirituality/religion, Western vs non-Western frame of reference). These themes are further stratified into a number of subthemes and one overarching theme (i.e., diversity policy). The themes have resulted in six recommendations to improve cultural psychiatric care. These recommendations underscore the vulnerability of the target patient population, specific training needs, the need for professional interpreters and intercultural mediators, the place of religion and spirituality in therapy, reflexivity as core competence, and the need to establish reference centers.

The six recommendations provide a scientifically sound base to develop focused and effective mental health policies at the governmental, organizational, and patient level. Continued attention to the importance of cultural sensitivity in mental healthcare provision remains important, particularly in countries that are lagging behind.

Parental alcohol supply in early childhood may increase the risk of alcohol use in late adolescence. This study examined its longitudinal impact and the distinct roles of mothers’ and fathers’ drinking.

We studied 1,891 mother–child pairs from the Czech European Longitudinal Study of Pregnancy and Childhood. Mothers reported parental alcohol supply at ages 3, 5, 7, and 11 years, while adolescent alcohol use was reported by mothers, pediatricians, and youth at ages 18 and 19 years. Structural equation modeling assessed the longitudinal link between early alcohol supply (three classes: none, occasional, and frequent) and adolescent alcohol use, accounting for parental drinking and covariates, including the child’s sex, mother’s education, and family structure.

Alcohol supply began in early childhood, with 14% of children exposed by age 3 and around 20% by age 11. By age 19, one-third of individuals reported frequent alcohol use. Adolescents’ alcohol use was associated with concurrent mothers’, but not fathers’ alcohol use (β = .24, p < .001). Early alcohol supply predicted higher adolescent use for both occasional (β = .14, p = .041) and frequent (β = .22, p = .005) classes. Mothers’ and fathers’ alcohol use at 6 months was associated with frequent alcohol supply, and fathers’ alcohol use was also associated with occasional alcohol supply. Significant indirect effects were found from early parental drinking to adolescent use via these classes.

Public health messaging should emphasize the risks of early alcohol consumption, including its potential harm to the developing brain.

The awareness of climate change as a global environmental threat through media consumption and/or social interaction can have a psychological impact on people’s mental health. However, little is known about the association between climate change awareness-related psychological distress (CCARPD) and mental health in people with psychiatric diagnoses or subclinical symptoms.

A comprehensive and systematic literature search of the PubMed, Embase, Web of Science Core Collection, Scopus, and CENTRAL electronic databases (from inception to February 2025) was conducted, without language restriction, for articles assessing the association between CCARPD and the mental well-being of people in the general and psychiatric populations.

Twenty-eight thousand forty-seven reports were retrieved. Of these, 67 met the inclusion criteria (64 general and 3 psychiatric population studies). The overall correlation between CCARPD and mental health measurements (ranging from subclinical symptoms to clinical diagnoses of depression, anxiety, or stress) was positive and of weak-to-moderate strength. Nevertheless, higher psychological distress due to the awareness of climate change was found in those having more severe mental health problems.

Although most studies have found small-to-moderate correlations between CCARPD and mental health measurements, it can be distressing and damaging for those with more severe mental health problems. As CCARPD will increase globally as the climate crisis unfolds in the coming decades while the understanding of the connections between CCARPD and mental well-being is still at an early stage of development, more research will be of utmost relevance, particularly in psychiatric populations.

Trazodone is commonly used in the treatment of major depressive disorder (MDD) in adults. This study aimed to establish consensus on the clinical scenarios and patient profiles for which trazodone treatment is considered suitable.

A two-round Delphi process was conducted across eight European countries. Statements regarding trazodone were rated by a panel of 32 experts for agreement or disagreement using a 9-point Likert scale; those with <70% agreement among panelists were revised and reassessed by the panel.

There was strong consensus agreement on 68 out of 91 statements (75%) related to trazodone. According to the panel, trazodone is well tolerated, with low anticholinergic activity, minimal impact on sexual function, weight neutrality, and low potential for clinically relevant drug–drug interactions. Consensus agreement supported trazodone use across a broad spectrum of patients with MDD, including those with insomnia, anxiety, psychomotor agitation, substance use, physical comorbidities, neurological conditions, and treatment-resistant depression; consensus agreement was also achieved for trazodone use in elderly patients, and those experiencing adverse effects with other antidepressants.

This study suggests that trazodone is useful in the treatment of MDD across multiple patient profiles. These findings offer practical guidance to support individualized and evidence-based decision-making in clinical practice.

The early phases of severe mental disorders are often diagnostically challenging, with frequent diagnostic shifts over time. Few studies have combined detailed baseline diagnostic assessment with long-term follow-up to examine both diagnostic and social development.

We conducted a 20-year register-based follow-up of 150 patients with first-time psychiatric hospitalizations. At baseline, all participants underwent a comprehensive diagnostic assessment. Follow-up data were obtained through linkage to national registers, providing information on psychiatric diagnoses, education, family formation, crime, mortality, and suicide. Cumulative incidence functions accounting for competing risks were calculated stratified on baseline diagnoses.

Only seven participants (4.6%) had no further contact with hospital-based psychiatry during the 20-year follow-up. During the follow-up period, 37.9% received a diagnosis of schizophrenia, 35% schizotypy, 14.4% depression, 24.6% personality disorder, 11% bipolar disorder, and 6.1% substance use disorder. Participants with a baseline diagnosis of schizophrenia, schizotypy, or depression had a significantly higher probability of receiving the same diagnosis during follow-up (schizophrenia 81.6%, schizotypy 69.4%, and depression 53.3%), whereas this was not the case for participants with a baseline diagnosis of personality disorder. Mortality was elevated (5.9%), with suicide accounting for one-third of all deaths, ten times the national average.

A first psychiatric hospitalization in early adulthood marked the beginning of a longer clinical trajectory: 95% of participants re-entered hospital-based care or had prolonged initial hospitalization. The findings emphasize the importance of diagnostic assessment and sustained care to improve prognosis and reduce social impairment and premature death.

Attention deficit hyperactivity disorder (ADHD) affects 5–7.2% of children and 2.5% of adults. Despite its prevalence, ADHD remains underdiagnosed and undertreated, leading to significant challenges for affected individuals. Early diagnosis and intervention can prevent adverse outcomes and improve quality of life.

We developed a predictive model to identify adults with ADHD using electronic health records. The dataset comprised 2,973 adult patients (aged 18 years and above) diagnosed with ADHD and a control group of 4,447 adults referred to psychologists with no ADHD diagnosis. A transformer-based architecture was implemented, utilizing only clinical codes and gender as input features. Fivefold cross-validation was adopted, and model performance was evaluated on held-out test data consisting of 800 patients, 400 of whom had an ADHD diagnosis.

Our study demonstrated the ability to predict adult ADHD using clinical data, with a 6-month model achieving an area under the receiver operating characteristic curve (AUC) of 0.79 (95% confidence interval: 0.76–0.81), F1-score of 0.79, sensitivity of 0.80, and specificity of 0.77. Shapley Additive Explanations identified key contributing codes, including F158 and Y903, consistent with known associations between ADHD and substance use.

Our findings show that machine learning can effectively use clinical codes and demographic data from routine EHRs to support early, cost-efficient diagnosis of adult ADHD, paving the way for earlier intervention and improved outcomes.

The prevalence of mood, anxiety, and substance use disorders has risen in the last decade. It is unclear to what extent this rise is also seen in the first-incidence of these disorders, even though this is relevant for prevention. We provide up-to-date information on the first-incidence of common mental disorders (mood, anxiety, and substance use disorders) and compare this with the first-incidence 12 years ago.

First-incidence of DSM-5 common mental disorders was examined with a slightly modified version of the Composite International Diagnostic Interview (CIDI) 3.0 in 4,688 respondents (18–75 years; interviewed in 2019–2022 and 2023–2024) from the third Netherlands Mental Health Survey and Incidence Study (NEMESIS-3). The CIDI also assessed DSM-IV diagnoses and, therefore, 12-year changes could be examined by comparing first-incidence rates of DSM-IV mental disorders between NEMESIS-3 (3,687 respondents aged 18–64 years) and NEMESIS-2 (5,303 respondents aged 18–64 years; interviewed in 2007–2009 and 2010–2012).

In NEMESIS-3, 11.1% of adults without prior psychopathology experienced a DSM-5 common mental disorder over 3 years. First incidence was similar for any mood disorder (7.1%) and any anxiety disorder (6.9%), but lower for any substance use disorder (3.2%). From 2010-2012 to 2023-2024, the 3-year incidence of any DSM-IV disorder significantly increased from 8.5 to 14.0%. This change remained significant after controlling for differences in sociodemographic characteristics.

The substantial rise in first incidence of mental disorders likely contributes to the previously observed rise in their prevalence. This implicates a need for enhanced preventive measures and early intervention initiatives.

Network analysis was employed to test whether the overall pattern of depressive–anxious symptom connections remains stable or whether specific symptom-to-symptom links shift from pregnancy to postpartum.

In a perinatal sample (n = 4,461 pregnant women, n = 5,711 postpartum women), depressive and anxiety symptoms were assessed with the Edinburgh Postnatal Depression Scale (EPDS) and Generalized Anxiety Disorder-7 (GAD-7). Phase-specific polychoric Gaussian graphical models were estimated with EBICglass. We examined strength and bridge centrality, community structure, and nodewise predictability, and compared networks using the network comparison test.

Depression and anxiety formed four reproducible communities (one GAD-7 worry/arousal and three EPDS affective/anhedonic, anxious–cognitive distress, and depressed affect/sleep–suicidality modules) with identical partitions across phases. Global strength was similar, but postpartum networks showed higher edge density and more negative partial correlations, suggesting localized changes in which symptom pairs were directly linked—and how strongly—across phases. Across phases, Sadness, Crying, Uncontrollable worrying, and Trouble relaxing were most central and predictable. Worry-, arousal-, and sleep-related symptoms (e.g., hard to sleep) showed the strongest bridge centrality postpartum, and Self-harm was a prominent bridge during pregnancy; several edges shifted between phases, including stronger Enjoyment–Self-harm and weaker Hard to sleep–Self-harm postpartum.

Perinatal depression and anxiety organize into cohesive yet partially distinct symptom networks that remain globally stable but show localized shifts in direct symptom-to-symptom connections from pregnancy to postpartum. Central affective and arousal nodes, particularly sadness, pathological worry, and sleep disturbance, may be high-yield targets for phase-tailored screening and intervention.